《證件照拍攝指南 On-Site》

年份2023
媒材：深偽技術與機器學習模型, 即時影像, 現場攝影機與網路監視器
尺寸：尺寸可變

延續與陳珊妮、談宗藩於2020年由狠主流媒體製作之行動式計畫，藉由IoT（Internet of Things）搜尋引擎，搜尋並利用未正確加密的網路攝影機或監視器去拍攝國民身分證件照，意圖藉由直接的行動去探討身份認證、資訊安全與主體辨識與認知之間的問題，本次政治大學藝文中心展覽，藉由臉與面容作為主題為出發點，藉由深偽技術去製作現場實體裝置，將網路監視設備與現場攝影裝置去置換被照者的面容。

Photo of ID_2023 On-Site

Year：2023
Material：deepfake script and ML models, realtime projection, cameras Size: size variable

Continuing the action produced by the Very Mainstream (狠主流) in 2020 involving Shandee Chen and TAN Tsung-Fan, an IoT (Internet of Things) search engine was used to search for and exploit improperly encrypted network cameras or monitors to capture national ID card photos. The aim was to directly explore the issues of identity authentication, information security, and subject recognition and cognition through direct action. This exhibition at the National Chengchi University Arts Center takes faces and facial features as its theme, and uses deepfake technology to create physical installations on site, replacing the subjects' faces with network surveillance and live camera equipment.

《The Face of Machine of Face》

年份2023
媒材：ChatGPT API 、微型電腦、螢幕、即時生成對話影像
尺寸：尺寸可變

藉由T5 text to text framework 去創造不同問句，在藉由ChatGPT API進行即時的思辯與回答，提問內容多與身份、面貌、機器需要人的臉或是面容與特徵等等問題。

The Face of Machine of Face

Year：2023
Material：ChatGPT API , computer, LCD monitor, realtime video content Size: size variable

The T5 text to text framework is used to create different questions that can be answered by the GPT API. ChatGPT API for real-time debates and answers to questions about identity, face, machine needs human face or face and features, etc.

《這可能是你》

年份2022
媒材：軟體、印表機、熱感紙，即時生成影像，裝置可變

作品源自2019年鄭先喻的個展｛同化者｝中無題作品之延伸，藉由網路伺服器端應用，搜尋並搜集地區性線上論壇與即時聊天室資訊，建立資料庫模型以及視覺產出。當網路鄉民一留言，系統運用發文或留言者的帳號去搜尋可能的個人資料與肖像照片，再藉由合成資料技術結合機器學習成像，產出一系列的假想個資。

觀眾將於現場觀看網路鄉民即時訊息，現場列印機裝置則定時將合成的個人資料列印產出。

圖片：臺灣當代實驗場CLab提供

It Could Be You

Year：2022
Material：Software, printer, thermal paper, real-time generative video, installations variable

As a continuation of the untitled works featured in CHENG Hsien-Yu's solo exhibition｛Assimilator｝ in 2019, It Could Be You employs internet servers to collect messages from regional online forums and chatrooms to create a database model and to generate visual images. When netizens post a message, the system will detect it and instantly search for any personal information and portraits available by the users' IDs, with which it adopts and integrates the technology of synthetic data with machine learning to generate images as a series of imaginary and fictional personal data.

Apart from the real-time and on-site display of the instant messages posted by netizens around the digital world, the installation will also print out the synthetic personal data at regular intervals.

《肖像2011》

年份2019
媒材：矽膠、金屬、塑膠、馬達、軟體、微控制器、感應器、水桶、廢棄物
尺寸: 依空間而定

開心、傷心或強烈情緒會讓人有哭泣的行爲，但有些人會將倔強、脆弱等多重情緒隱藏其後，不希望讓人看見背後深沉的一面，於是「在你背後哭泣」（Crying behind your back）成為這件作品最直觀的概念：展場中的機械人頭，會對觀眾產生細微的動作與表情的反應，但遇空無一人時，它便會低頭哭泣。

圖片：鄭先喻提供

Portrait 2011

Year：2019
Material：Silicone, metal, plastic, motor, software, MCU, sensors, water tank, junks
Size：Size variable

The installation is a robotic head. When there are people in the exhibition space, it will be move and react with some facial expressions, but when there is no one in the space, it will bow its head and cry.

Happy, sad or strong emotions which cause the biological behaviour, "cry", but some people will hide strong emotions behind people and do not want to be seen.